返回新闻
产品展示AI Understanding 简报

Microsoft 推出新的 Copilot 功能,包括自动驾驶仪和代码

Microsoft 正在更新其 Copilot 平台,提供三个核心功能——Home、Code 和 Autopilot——以及用于管理企业支出的 AI 工具的新 FinOps。

4 min readRead the linked source
Source-provided image accompanying Microsoft introduces new Copilot features including Autopilot and Code
来源参考来源记录
出版商
blogs.microsoft.com
来源链接
blogs.microsoft.comhttps://blogs.microsoft.com/blog/2026/09/25/introducing-the-new-copilot-with-home-code-and-autopilot/
来源类型
链接来源——主要来源状态尚未确定。
背景60 秒内了解这一点

从这里开始

关键术语

API(应用程序编程接口)
一种软件系统向另一个系统发送请求并接收响应的结构化方式。
内存(代理内存)
AI 代理跨步骤或会话使用存储的上下文来提高连续性。
基于使用的计费
成本随 API 调用、令牌、推理时间或消耗的计算而变化的定价。
测试一下自己AI 代理测验

发生了什么

Microsoft has announced a major overhaul of its Copilot platform, introducing three primary functional pillars: Home, a centralized interface for chat and task delegation; Code, a tool for building custom, sandboxed applications; and Autopilot, a persistent agent designed to perform autonomous, long-running tasks. The company is also integrating Microsoft Office applications directly into the Copilot experience and introducing 'FinOps for AI' to provide administrators with granular control over and model selection.

The new 'Home' interface serves as a unified hub for Chat and 'Cowork,' the latter being a feature for delegating end-to-end tasks like RFP responses or financial close packages. Microsoft is also embedding the full functionality of Word, Excel, and PowerPoint directly into the Copilot experience, allowing users to create and edit documents in real-time without switching applications.

The 'Code' capability allows users to build small, purpose-built solutions—such as desktop widgets or internal dashboards—using natural language. These solutions run in a sandboxed environment via the 'Microsoft Copilot Managed Runtime,' which is governed by IT and hosted within the company's tenant.

Autopilot, formerly known as 'Scout,' acts as a persistent digital teammate. Unlike standard chat-based AI, it is cloud-hosted and designed to operate autonomously, monitoring channels and managing recurring workflows even when the user is offline. It functions with its own identity and memory, grounded in the organization's data via 'Microsoft IQ.'

Microsoft is implementing a dual-pricing model: a fixed-cost User Subscription License (USL) for everyday tasks and (UBB) for more advanced, agentic capabilities like Cowork, Code, and Autopilot. New FinOps for AI tools allow administrators to set spending policies, manage API access, and restrict which model families are available to specific user groups.

来源详情: blogs.microsoft.com ↗

为什么这很重要

This update represents a shift from simple conversational AI to an agentic framework where the software can autonomously execute complex, multi-step workflows across an organization's data. By introducing 'Autopilot' and 'Code,' Microsoft is attempting to lower the barrier for non-developers to build custom business applications while simultaneously providing IT departments with the governance and cost-management tools necessary to scale these AI deployments within enterprise environments. The move signals a transition toward AI that operates independently of direct user prompts.

The integration of 'Code' and 'Autopilot' suggests a strategic push to make AI a primary interface for business operations rather than just a productivity assistant. By allowing non-technical staff to build their own tools, Microsoft is attempting to decentralize software development within the enterprise.

The introduction of FinOps for AI is a critical response to the unpredictability of enterprise AI costs. By providing admins with the ability to route credit requests and set model-specific usage limits, Microsoft is addressing a primary barrier to the widespread adoption of agentic AI in corporate settings.

The reliance on 'Microsoft IQ' and 'Fabric IQ' to ground these agents in enterprise data—including Power BI semantic models and Dynamics 365 workflows—highlights the company's strategy to leverage its existing data ecosystem to differentiate its AI offerings from competitors.

Interactive Mechanism

互动机制:它实际上是如何运作的

以交互方式探索这一发展背后的基础技术。

Agent Lifecycle Stage:
1
User Intent & Planning: "Audit customer refund request #4092 and settle payment."
2
Tool Calling: Emits structured JSON call crm_get_transaction(id='4092').
3
Guardrail & Verification:🛡️ Paused: High-value action requires human operator sign-off.
4
Final Settlement: Refund recorded, email receipt dispatched, and audit log stored.
Core takeaway: An AI agent is not just a language model—it is a closed loop of planning, tool invocation, and environment feedback. Production systems require self-healing retries and strict human approval guardrails.
交互式概念检查+10 Points
AI Agents Quiz

An agent must create a draft calendar event for Tuesday at 2 p.m. Which evidence would establish the requested result?

接下来看什么

The rollout schedule is staggered: Home and Code are entering the 'Frontier' program in the coming weeks, while Autopilot is scheduled for private preview at the end of September. Microsoft has also scheduled a preview for Microsoft 365 Premium and Pro subscribers later this year. Observers should monitor the actual performance of the 'Autopilot' agent in real-world enterprise environments, particularly regarding its ability to manage complex, multi-stakeholder processes without human intervention, and how effectively the new FinOps tools prevent cost overruns in large-scale deployments.

The effectiveness of the 'Autopilot' agent in managing complex, multi-step business processes remains an unknown. The ability of the system to maintain context and adhere to organizational governance while operating autonomously will be a key metric for enterprise adoption.

The transition to for agentic features may create friction for organizations accustomed to flat-rate software licensing. Monitoring how companies balance the value of these agents against the potential for variable, high-cost usage is essential.

Microsoft has announced that 'Today,' a proactive command center feature, will enter private preview in October. Its integration into Outlook and Teams will be a significant test of the company's ability to manage user attention without creating notification fatigue.

相关指南和测验

人工智能代理人工智能模型解释AI 的未来测试你所知道的——尝试免费的人工智能测验在我们的词汇表中查找人工智能术语关注 AI 模型发布跟踪器
觉得这有用吗?